Repair Technique For Image Coding Based On Pde Research

The technology of image coding plays a very important role in data storage and transmission, and its essential idea is to use the minimum amcunt of data to express a relatively complete information, some redundant information will be removed in encoding process. The core thought of image inpainting technique is to restore the lost information based on existing information by using a certain criteria and algorithm. From this perspective, image coding and inpainting techniques can unify together organically. The method of this paper is just based on this kind of consideration.In this paper, many classic inpainting algorithms based on partial differential equation and JPEG2000 are discussed. An improved solution aiming at solving the defects of G-C model was given.Then, an image coding system combining image inpainting technology with JPEG2000 was designed. The main jobs of this paper are shown as following:Firstly, the paper compares the principle algorithms and numerical solutions of BSCB model, TV model and CDD model, which are all classical PDE inpainting algorithms. The inpainting effects of each algorithm are verified with simulation experiments.Secondly, the improved G-C model is proposed. Edge stopping function of G-C inpainting model is reconstructed to remedy the existing problems of edge stoping function. Experiments indicate that the improved G-C model has higher repair efficiency and better repairing effect than original G-C model.Finally, a coding system based on G-C inpainting model and JPEG2000 was designed. Based on the analysis of the rational scheme of the system, detailed analysis of functions of each module and algorithm is given. Experiment results proved that, the quality of reconstructed image is better than JPEG2000 coding system in high compression percentage cases; Compared to the coding system based on P-Harmonic model, the compression performance also have certain degree of improvement.